metadata
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- wmt14
metrics:
- bleu
model-index:
- name: t5-small-finetuned-de-en-epochs5
results:
- task:
name: Sequence-to-sequence Language Modeling
type: text2text-generation
dataset:
name: wmt14
type: wmt14
args: de-en
metrics:
- name: Bleu
type: bleu
value: 8.2154
t5-small-finetuned-de-en-epochs5
This model is a fine-tuned version of t5-small on the wmt14 dataset. It achieves the following results on the evaluation set:
- Loss: 2.0694
- Bleu: 8.2154
- Gen Len: 17.3996
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Bleu | Gen Len |
---|---|---|---|---|---|
No log | 1.0 | 375 | 2.0923 | 8.0197 | 17.4046 |
2.3091 | 2.0 | 750 | 2.0806 | 8.0314 | 17.4186 |
2.2602 | 3.0 | 1125 | 2.0746 | 8.1423 | 17.4033 |
2.2337 | 4.0 | 1500 | 2.0702 | 8.2025 | 17.4029 |
2.2337 | 5.0 | 1875 | 2.0694 | 8.2154 | 17.3996 |
Framework versions
- Transformers 4.12.5
- Pytorch 1.10.0+cu111
- Datasets 1.16.1
- Tokenizers 0.10.3